JP2013104751A - Incidence condition determination method, database creation method, pattern recess thickness measurement method, imprint device and article manufacturing method - Google Patents

Abstract

A technique advantageous in measuring the thickness of a concave portion of a pattern is provided.
A method of determining an incident condition of light used for calculating a remaining film thickness is a light that generates reflected light in a first polarization state in which a difference in measurement sensitivity between the remaining film thickness RLT and a pattern thickness is large. Obtained for each of a plurality of combinations of the plurality of first incident conditions and a plurality of second incident conditions for light that generates reflected light in the second polarization state with good measurement sensitivity of the remaining film thickness RLT. Create a regression equation with the intensity of the reflected light as an independent variable and the thickness of the recess obtained using the information as a dependent variable, and a regression error satisfying an acceptable condition among the created regression equations A first incident condition corresponding to the equation as an incident condition for generating the reflected light in the first polarization state, and a second incident condition corresponding to the regression equation as an incident condition for generating the reflected light in the second polarization state, Decide each.
[Selection] Figure 1

Description

  The present invention relates to an incident condition determination method, a database creation method, a pattern concave thickness measurement method, an imprint apparatus, and an article manufacturing method.

  The imprint technology enables mass production at low cost without requiring a large-scale apparatus such as a vacuum process. Therefore, it is expected to be applied as a technique for forming a two-dimensional fine pattern with high accuracy in place of conventional photolithography in next-generation semiconductor processes. According to the technology strategy roadmap formulated by the Ministry of Economy, Trade and Industry, 2013 (32 nm technology node) is listed as the target year for applying imprint technology to semiconductor processes. Application of imprint technology to semiconductor processes is positioned as one of the specific industrial needs in the next-generation semiconductor industry.

  Generally, in a resist pattern formed by an imprint technique, a thin resist film having a thickness of about several tens of nanometers remains in the recess as an ultrathin residual film. Therefore, in the semiconductor process, a step of removing this remaining film is required. If the remaining film varies, the dimension of the resist pattern after removal of the remaining film by etching varies. Then, since the resist pattern after removal of the remaining film functions as an etching mask for the work-in-process of the semiconductor device, it has been regarded as a problem that the characteristics of the semiconductor device vary and cause defects. In order to solve such problems and apply the imprint technology to the semiconductor process with high reliability, it is necessary to establish a technology for accurately controlling the remaining film thickness of about several tens of nanometers generated during imprinting. is there. For that purpose, highly accurate measurement of the residual film thickness of the resist is indispensable. Patent Document 1 describes measuring the remaining film thickness and pattern dimensions, but does not disclose a specific method for measuring the remaining film thickness. In semiconductor manufacturing, oxide films, nitride films, and metal films are used as semiconductor device structures, and there are many methods for measuring the film thickness. In particular, measurement methods using ellipsometry are widely used. .

JP 2010-030153 A

  When measuring the cross-sectional shape of the concavo-convex pattern using the ellipsometry method to measure the residual film thickness formed by imprinting, the height of the concavo-convex pattern and the residual film thickness of the resist can be measured separately. The problem of difficulty arises. Specifically, the problem will be described with reference to FIG. FIG. 4A is a diagram illustrating a concavo-convex pattern formed by the imprint process. 4B and 4C are diagrams in which only the resist portion is extracted. When the total height h of the resist portion is the same, if the residual film thickness (Residual Layer Thickness: RLT) changes from FIG. 4B to FIG. 4C, the pattern height (Height: HT) also changes accordingly. To do. The pattern portion and the remaining film portion of the resist portion are the same resist resin. Therefore, their optical constants (refractive index n, absorption coefficient k) are the same, and it is assumed that the output (spectral characteristics) obtained by measuring the resist patterns of FIGS. 4B and 4C by the ellipsometry method is not significantly different. Therefore, it is difficult to distinguish whether the remaining film thickness has changed or the pattern height has changed from the result of the spectral characteristics.

  This will be further described using the results of simulation with an optical simulator (for example, RCWA) of a vector diffraction model in which the remaining film thickness (RLT) and the pattern height (HT) are variously changed in the cross-sectional shape of the pattern. FIG. 5A shows the result of spectral characteristics when RLT and HT are variously changed when the measurement condition is 0 degree polarization. FIG. 5B shows the relationship between the difference from the nominal intensity value and the wavelength, and changes due to HT and changes due to RLT are mixed. In the region indicated by the region A having a wavelength region from 0.5 μm to a high wavelength region, there are a plurality of combinations of RLT and HT that indicate the same spectral characteristic output (intensity difference). FIG. 5C is a schematic diagram in the case where the horizontal axis is RLT and HT is a parameter from FIG. 5B. In the examples of FIGS. 5B and 5C, when the combination of RLT and HT is (RLT1, HT1), (RLT2, HT2), (RLT3, HT3), the same intensity difference is shown in region A. Therefore, the RLT cannot be specified from the spectral characteristic output. Patent Document 1 does not disclose a specific method for measuring the remaining film thickness, and does not point out a problem that the pattern height and the resist remaining film thickness cannot be separated.

  The present invention has been made in view of the above circumstances, and an exemplary object thereof is to provide a technique advantageous for measuring the thickness of a concave portion of a pattern formed by imprinting.

  The present invention includes a measurement result obtained by measuring the intensity of reflected light obtained from the pattern with light obliquely incident on the pattern, and information indicating a relationship between the intensity of the reflected light and the shape of the pattern. A method of determining an incident condition of the light used to determine a thickness of the concave portion of the pattern, and using the reflected light of the first polarization state from the pattern with respect to the thickness of the concave portion The measurement with respect to the differential thickness corresponding to the difference between the thickness of the convex portion of the pattern and the thickness of the concave portion measured using the reflected light in the first polarization state, with the measurement accuracy of the measuring instrument being the first accuracy. The measurement accuracy of the measuring device is the second accuracy, the measurement accuracy of the measuring device with respect to the thickness of the recess measured using the reflected light of the second polarization state from the pattern is the third accuracy, and the second polarization state Measurement using reflected light The measurement accuracy of the measuring instrument with respect to the differential thickness is set as the fourth accuracy, the measurement accuracy is improved in the order of the second accuracy, the first accuracy, and the third accuracy, and the first accuracy and the The first polarization state and the second polarization state are set so that the difference from the second accuracy is larger than the difference between the third accuracy and the fourth accuracy, and reflected light in the first polarization state is generated. Combining one of a plurality of first incident conditions as the incident condition for the purpose and one of a plurality of second incident conditions as the incident condition for generating the reflected light in the second polarization state For each of the plurality of combinations, a regression equation is created using the obtained reflected light intensity as an independent variable and the thickness of the recess obtained using the information as a dependent variable. Regression where the regression error satisfies the allowable condition As the incident condition for generating the reflected light in the first polarization state, the second incident condition corresponding to the regression equation as the incident condition for generating the reflected light in the second polarization state, respectively. It is characterized by determining.

  ADVANTAGE OF THE INVENTION According to this invention, the technique advantageous to the measurement of the thickness of the recessed part of the pattern formed by the imprint can be provided, for example.

The flowchart which shows the measuring method of 1st Embodiment. Configuration diagram of imprint device Control block diagram of imprint device The figure which shows the cross-sectional shape of the uneven | corrugated pattern formed with an imprint apparatus The figure which shows the spectral characteristic by the reflected light of 0 degree polarization Diagram showing the optical system of the measuring instrument The figure which shows the measuring method of the cross-sectional shape of an uneven | corrugated pattern Diagram showing the accuracy of RLT and HT The figure which shows the spectral characteristic by the reflected light of 45 degree polarization Diagram showing how to measure the remaining film thickness The figure which shows the shot layout of 3rd Embodiment Flowchart showing the third embodiment Flowchart showing the sixth embodiment Diagram showing how to create a regression equation It is a figure showing the prediction measurement precision of RLT when noise is added. It is a figure showing the prediction measurement precision of RLT when noise is added. It is a figure showing the prediction measurement precision of RLT when noise is added.

[First Embodiment]
A measurement method according to the first embodiment will be described. FIG. 2 is a diagram illustrating an example of the configuration of the imprint apparatus, and FIG. 3 is a diagram illustrating an example of a control block of the imprint apparatus. 2 and 3, the substrate (wafer) 1 is held by a wafer chuck 2. The fine movement stage 3 has a function of correcting the rotation of the wafer 1 around the z axis, adjusting the z position of the wafer 1, and correcting the tilt of the wafer 1, and is disposed on the XY stage 4. The fine movement stage 3 and the XY stage 4 are collectively referred to as a wafer stage. An XY stage 4 is placed on a base surface plate 5, and a reference mirror (not shown) that reflects light from a laser interferometer in the x and y directions on the fine movement stage 3 in order to measure the position of the fine movement stage 3. Is attached.

  A concave / convex pattern to be transferred to the wafer 1 is formed on the surface thereof, and a mold (mold) 10 for molding the photo-curing resin (resist) supplied on the wafer 1 is molded mold chuck by a mechanical holding mechanism (not shown). 11 is fixed. The mold chuck 11 is placed on the mold chuck stage 12 by a mechanical holding mechanism (not shown). The mold chuck 11, the mechanical holding mechanism (not shown), and the mold chuck stage 12 constitute a support that supports the mold (mold 10). The mold chuck stage 12 has a function of correcting the rotation of the mold 10 (mold chuck 11) about the z axis and correcting the tilt of the mold 10. The mold chuck stage 12 has a mirror reflection surface (not shown) that reflects light from the laser interferometer in order to measure the position of the mold 10 in the x and y directions. The mold chuck 11 and the mold chuck stage 12 each have an opening (not shown) through which the UV light irradiated from the UV light source 16 through the collimator lens passes to the mold 10. One end of the guide bar plate 13 is fixed to the mold chuck stage 12 and the other end of the guide bar 14 penetrating the top plate 9 is fixed. The mold raising / lowering linear actuator 15 is composed of an air cylinder or a linear motor, and drives the guide bar 14 in the z direction to press the mold 10 held by the mold chuck 11 against or away from the wafer 1. The guide bar 14 penetrates the alignment shelf 18 suspended from the top plate 9 by the support column 19.

  A die-by-die alignment TTM (through-the-mold) alignment scope 20 has an optical system and an imaging system for observing alignment marks provided on the wafer 1 and the mold 10. The TTM alignment scope 20 measures the positional deviation between the wafer 1 and the mold 10 in the x and y directions. The alignment shelf 18 is provided with a height measuring instrument (not shown) for measuring the height (flatness) of the wafer 1 on the wafer chuck 2 by using, for example, an oblique incident image shift method. The dispenser head (supply unit) 30 includes a resin dropping nozzle that drops a liquid photocurable resin (resist) on the surface of the wafer 1. The CPU (control unit) 40 controls the above-described actuators and sensors to cause the imprint apparatus 100 to perform a predetermined operation. The imprint apparatus 100 includes a memory 50. The imprint apparatus is connected to the measuring instrument 300 and the control apparatus 400 via the communication unit 500. The control device 400 is connected to the imprint apparatus 100 and the measuring instrument 300 via the communication unit 500, controls the entire semiconductor process, and feedback-controls information on the remaining film thickness from the measuring instrument 300 to the imprint apparatus 100. . A feedback control method will be described in an embodiment described later.

  The measuring instrument 300 has a thickness (remaining film thickness: RLT) of the concave / convex pattern of the resist formed on the wafer 1 and a difference thickness (pattern height: HT) between the thickness of the convex and the concave. Measure. The measuring device 300 is a device that measures a cross-sectional shape using an ellipsometry method, and is commercially available from a measuring instrument manufacturer as a device that measures the CD (Critical Dimension), that is, the line width of an uneven pattern, using light. The measuring instrument 300 can measure not only the CD measurement but also the remaining film thickness (RLT), the height of the concavo-convex pattern (HT), the side wall angle (Side wall angle), and the like. In the present embodiment, the measurement device 300 is used to inspect the concavo-convex pattern of the resist formed on the wafer 1 to measure RLT and HT. The measuring instrument 300 will be described with reference to FIGS. FIG. 6 shows an example of an optical system of a measuring instrument 300 that measures the cross-sectional shape of the concavo-convex pattern using the ellipsometry method. The ellipsometry method includes a method of spectrally detecting the wavelength of reflected light when a broadband light including light of a plurality of wavelengths is obliquely incident at a constant incident angle and a single wavelength of light having a plurality of incident angles. There is a method of spectrally detecting the wavelength of reflected light when obliquely incident. In the present embodiment, the former method is used, but the latter method can also be used. Such a method of spectrally detecting the wavelength of the reflected light is called a spectroscopic ellipsometry method. The light 4 a emitted from the light source 41 passes through the rotatable polarizer 42, so that the polarization plane (S-polarized light, P-polarized light) is adjusted and the phase is adjusted, and the light 4 a enters the concavo-convex pattern 43. The light 4b reflected by the concave / convex pattern 43 is spatially wavelength-separated by the spectroscopic optical system 44 according to the wavelength. In the separated light 4c, the intensity ratio and phase difference of S-polarized light and P-polarized light for each wavelength are detected by a photodetector 45 in which photoelectric elements are arranged in an array, and the information is sent to a computer 46. The computer 46 compares the information from the light detector 45 with library information described later to calculate the cross-sectional shape of the concave / convex pattern 43, and the calculation result is output from the computer 46 as the cross-sectional shape of the concave / convex pattern 43 as a measurement object. Is done. In the spectroscopic ellipsometry method, even if the resist film thickness and the resist pattern shape are different, the same intensity ratio / phase difference state may occur depending on the incident conditions of incident light (incident angle, wavelength of incident light, etc.). Therefore, measurement accuracy is also improved by detecting a change in reflected light under a plurality of incident conditions (irradiation with a plurality of incident angles or irradiation with incident light with a plurality of wavelengths).

  Next, a method for measuring the cross-sectional shape of the concavo-convex pattern 43 as a measurement object will be described with reference to FIG. First, as a preparatory stage before measurement, the assumed cross-sectional shape of the concavo-convex pattern 43 is defined on the computer 46 using an optical simulator of a vector diffraction model. The vector diffraction model is, for example, RCWA (Rigorous Coupled Wave Analysis), and is a model that solves an eigenvalue equation by expressing an electric field in a periodic structure by Fourier transform. The computer 46 obtains the state of reflected light from each cross-sectional shape defined next by simulation calculation, and stores a set of data in that state as a database (library). That is, the calculator 46 acquires the optical constants (refractive index n, absorption coefficient k, thickness d of each substance) of the substances constituting the concave / convex pattern 43 in step 1, and in step 2, the incident condition (wavelength λ or The light beam is incident on the concave / convex pattern 43 by changing the incident angle θ). Next, the computer 46 assumes various cross-sectional shapes of the concavo-convex pattern 43 in step 3, and information (intensity ratio change and level) obtained from reflected light when incident light is incident on the concavo-convex pattern 43 in step 4. Change in phase difference) is calculated. The calculator 46 performs this series of calculations on various defined different cross-sectional shapes, and associates the obtained calculation results with the corresponding cross-sectional shapes and stores them as a library. Here, the library refers to data associated with the cross-sectional shape or optical constant of the concave-convex pattern 43 corresponding to the light flux state obtained by simulation calculation based on optical constants and concave-convex patterns having various cross-sectional shapes, or Refers to the database. This stored library is information calculated by simulation showing the relationship between the variation in the intensity of reflected light with respect to the change in wavelength or incident angle and the shape of the concavo-convex pattern 43.

  The optical constants of the concavo-convex pattern used in step 1 differ depending on whether the incident angle θ or the incident light wavelength λ is used as the incident condition parameter. When the incident angle θ is changed using a single wavelength and measurement is performed using reflected light obtained for each incident angle θ, an optical constant for the single wavelength used for measurement is input to the computer 46. On the other hand, when the incident angle θ is fixed and the wavelength λ of the incident light is changed and the reflected light obtained for each wavelength λ is measured, the optical constant for each wavelength used for the measurement needs to be input to the computer 46. There is. Next, in step 5, using the measuring instrument 300, the incident light 4a is actually incident on the concave / convex pattern 43 as a measurement object, and in step 6, information on the reflected light 4b obtained (change in intensity ratio, level). Change in phase difference) is detected by the photodetector 45. In step 7, the computer 46 compares the information on the actual reflected light 4b from the library by comparing the light flux information of the library stored in the storage device with the information on the actual reflected light 4b obtained from the detected value. Is extracted. In step 8, the computer 46 determines the cross-sectional shape of the concave / convex pattern 43 associated with the light flux information among the cross-sectional shapes defined on the computer 46 as the actual cross-sectional shape of the concave / convex pattern 43. In the present embodiment, the measuring instrument 300 uses a spectroscopic ellipsometry method that changes the wavelength as an incident condition, and information of reflected light obtained for each wavelength λ is referred to as spectral characteristics.

  Next, a method for measuring the remaining film thickness of the resist will be described with reference to FIG. In S100, the computer 46 of the measuring instrument 300 obtains the measurement accuracy of the remaining film thickness RLT and the measurement accuracy of the pattern height HT while changing the polarization state of the incident light with respect to the concave / convex pattern 43 that is the measurement object. Here, the measurement accuracy will be described. The minimum CD resolution that can be measured by the measuring instrument 300 used in current optical lithography is ΔCD = 0.2 nm. First, an evaluation criterion represented by the following formula, that is, a value S (= Σ |) obtained by integrating the absolute value of the difference between the wavelengths of the two spectral characteristics I1 and I2 having a CD value difference of 0.2 nm in the wavelength region. I2 (λ) −I1 (λ) |) is defined. Next, when the RLT is changed instead of the CD, an RLT difference (ΔRLT) that takes the same value as S is calculated in a plurality of polarization states. On the other hand, when only HT is changed, a difference in HT (ΔHT) that takes the same value as S is calculated in a plurality of polarization states. In the present specification, ΔRLT and ΔHT are defined as measurement accuracy in the polarization state. A small value of the measurement accuracy means that the measurement accuracy is good. An example of the result of calculating the measurement accuracy is shown in FIG.

  The measurement by the measuring instrument 300 will be described with reference to FIG. The polarizer (polarizer) 42 converts the light from the light source 41 into linearly polarized light, and causes the converted linearly polarized light to enter the concavo-convex pattern 43. In addition, an analyzer (not shown) is present in front of the spectroscopic optical system 44, and transmits only a polarized light component having a predetermined polarization direction out of elliptically polarized light reflected by the uneven pattern 43. The predetermined polarization directions are, for example, 0 degree, 45 degrees, 90 degrees, and 135 degrees. The photodetector 45 receives the polarization component of a predetermined polarization direction transmitted by the analyzer. In FIG. 8, for example, 0 degree polarization means that the polarization direction of reflected light is 0 degree.

The inventor of the present application has found that there is a difference between the measurement accuracy of HT and the measurement accuracy of RLT depending on the measurement conditions. For example, when the four polarization states shown in FIG. 8 are compared, the following can be understood.
(1) In any polarization state, ΔRLT is smaller than ΔHT, that is, the measurement accuracy of RLT (remaining film thickness) is better than the measurement accuracy of HT (pattern height).
(2) In the polarization state of 45 degrees, the difference between ΔHT and ΔRLT, that is, the difference between the measurement accuracy of RLT (remaining film thickness) and the measurement accuracy of HT (pattern height) is the largest.
(3) In the polarization state of 0 degree, ΔRLT is the smallest, that is, the measurement accuracy of RLT (residual film thickness) is the best.

  The step of setting the first polarization state and the second polarization state for measuring the RLT based on the difference between the HT measurement accuracy and the RLT measurement accuracy is S110. In S110, the computer 46 sets the first polarization state and the second polarization state in which the measurement accuracy of RLT (residual film thickness) and the measurement accuracy of HT (pattern height) satisfy the following conditions. In the first polarization state, RLT measurement accuracy (first accuracy) is better than HT measurement accuracy (second accuracy), and RLT measurement accuracy (first accuracy) and HT measurement accuracy (second accuracy) Is a polarization state satisfying ΔRLT << ΔHT. In such a polarization state, since the measurement accuracy of HT is extremely lower than the measurement accuracy of RLT, the same spectral characteristics are exhibited as long as the RLT value is the same even if the HT value is different. Therefore, RLT can be obtained by separating the contribution of HT and the contribution of RLT from the measurement result in the first polarization state. In the second polarization state, RLT measurement accuracy (third accuracy) is better than RLT measurement accuracy (first accuracy) in the first polarization state, but RLT measurement accuracy (third accuracy) and HT measurement accuracy (first accuracy). (4 accuracy) is a polarization state in which the difference from the first polarization state is smaller. Since the difference between RLT and HT measurement accuracy is small in the second polarization state, the contribution of RLT and the contribution of HT cannot always be separated. Therefore, the RLT value cannot necessarily be uniquely determined from the measurement result of the second polarization state. However, the RLT value obtained from the measurement result in the second polarization state is selected from the RLT values obtained from the measurement result in the first polarization state, thereby obtaining a more accurate RLT value. I can do it. In the present embodiment, the 45-degree polarization state with the largest difference between ΔHT and ΔRLT corresponds to the first polarization state, and the 0-degree polarization state with the smallest ΔRLT corresponds to the second polarization state. The condition selected from the wavelength and incident angle of the light that generates the reflected light in the first polarization state is the first incident condition, and is selected from the wavelength and incident angle of the light that generates the reflected light in the second polarization state. The condition is a second incident condition. The method for obtaining the two polarization states in S110 has been described above.

  In S120, the intensity of the reflected light is measured by the measuring instrument 300 for each of a plurality of wavelengths of the reflected light using the reflected light in the first polarization state. In S130, the computer 46 narrows down the RLT with the first accuracy based on the measurement result in S120 and the library information. In the present embodiment, the first polarization state is a case of 45 degree polarization, and since HT is much less sensitive than the RLT in the 45 degree polarization state, it is possible to limit the RLT. explain. FIG. 9 shows the results of spectral characteristics simulated by an optical simulator (for example, RCWA) of a vector diffraction model in the case of 45 degree polarization which is the first polarization state. In FIG. 8, in the case of 45 degree polarization, the difference between the RLT measurement accuracy (ΔRLT45) 0.35 and the HT measurement accuracy (ΔHT45) 1.9 is the largest. Using ΔRLT45 and ΔHT45 having the largest difference in measurement accuracy, the RLT value is changed to −2ΔRLT45, −ΔRLT45, 0, + ΔRLT45, and + 2ΔRLT45 in the order from FIG. 9A to FIG. 9E in the difference from the nominal value. The calculation result of the intensity difference of the spectral characteristics is shown. The five graphs shown in each of FIGS. 9A to 9E show that when the value of RLT is the same, the value of HT changes from the nominal value to −2ΔHT45, −ΔHT45, 0, + ΔHT45, and + 2ΔHT45. The calculation result is shown. Referring to FIG. 9, it can be seen that the change in HT is insensitive to the change in RLT in the wavelength range from 0.4 μm to a high wavelength range. Regardless of the value of HT, the value of RLT can be narrowed down to the vicinity of a certain RLT. Thus, it was found that using the measurement result of 45-degree polarized light may limit RLT without being affected by HT. As described above, the RLT value is narrowed down using the reflected light in the first polarization state.

  In S140, the measuring device 300 measures the intensity of each of the plurality of wavelengths of the reflected light using the reflected light in the second polarization state. In S150, the computer 46 calculates the RLT with the third accuracy based on the measurement result in S140 and the library information. Referring to FIG. 8, the RLT measurement accuracy (ΔRLT0) in the case of 0-degree polarization in the second polarization state is 0.14 nm, which is the measurement accuracy in the case of 45-degree polarization in the first polarization state ( ΔRLT45) Higher than 0.35 nm. FIG. 10A is a schematic diagram in which the horizontal axis is RLT and HT is a parameter with respect to the spectral characteristic result of 45 degree polarized light which is the first polarization state of FIG. Further, FIG. 10B is a schematic diagram in which the horizontal axis is RLT and HT is a parameter with respect to the spectral characteristic result measured with 0-degree polarized light that is the second polarization state. For the purpose of measuring RLT with high accuracy, it is desirable to use the measurement result of the polarization state with higher measurement accuracy of RLT. In the present embodiment, as shown in FIG. 10A, the RLT value range is narrowed down to the region C from the measurement result (intensity difference = B) of the first polarization state (45-degree polarization). However, since the measurement accuracy of 45-degree polarized RLT is 0.35 nm, RLT cannot be measured with higher accuracy.

  In device manufacturing, the measurement accuracy of RLT (residual film thickness) is required to be suppressed to 0.2 nm or less, and the required accuracy is not satisfied only from the measurement result of 45-degree polarized light that is the first polarization state. Accordingly, in the present embodiment, measurement is performed using reflected light in the second polarization state (0-degree polarization) in S140, and three RLT candidate values (in FIG. 10B) whose intensity difference shown in FIG. The RLT in the region C is selected from the three white circles) to determine the RLT. That is, RLT can be measured with high accuracy by using the measurement result of the second polarization state (0 degree polarization) with higher measurement accuracy.

[Second Embodiment]
In the first embodiment, the RLT measurement of the resist is described as being performed inside the measuring instrument 300. However, the RLT measurement of the resist is not limited to being performed by the computer 46 inside the measuring instrument 300. For example, the control device 400 outside the measuring instrument 300 in FIG. 2 performs RLT measurement by exchanging necessary data from the measuring instrument 300 via the communication unit 500 and executing a predetermined program. May be.

[Third Embodiment]
The third embodiment is characterized in that the variation between shot areas in a wafer is calculated for each lot regarding the RLT of a resist. Specifically, the measuring device 300 measures the pattern shape in 16 shot areas (locations) as shown in FIG. Is evaluated by 3σ or the like. It is possible to determine whether the RLT of the lot is stable based on variations in RLT. Further, the third embodiment is characterized in that the imprint apparatus 100 is controlled based on the variation of RLT.

  FIG. 12 is a diagram showing a flow when RLT measurement is performed for each lot. In S200, the wafer of the first lot is introduced into the imprint apparatus 100, and in S210, all wafers (for example, 25 sheets) in the lot are imprinted using the imprint apparatus 100. The imprint process includes alignment, resist coating, imprinting, curing, and releasing of each wafer. In S230, the computer 46 inside the measuring instrument 300 measures RLT in a plurality of shot areas for a specific number of wafers in the lot. In S240, the computer 46 calculates and evaluates a variation between RLT shot areas. If the variation exceeds a predetermined threshold value in S250, the computer 46 changes the recipe condition for the imprint apparatus 100 in S260. In the third embodiment, the supply amount of resist (ultraviolet curable resin) is changed as a recipe to be changed. For a shot region with a large RLT, the CPU 40 (control unit) changes the resist supply amount to be small. Conversely, for shot areas with a small RLT, the CPU 40 changes so as to increase the resist supply amount and adjusts so that variations between RLT shot areas are reduced. The control device 400 determines whether all lots have been completed in S270, and if not completed, introduces the next lot in S280 and performs imprint processing according to the recipe changed in S260.

[Fourth Embodiment]
In the third embodiment, in S260, the resist supply amount in the imprint apparatus 100 is changed based on variations between RLT shot areas. In the fourth embodiment, the CPU 40 (control unit) controls the gap between the wafer 1 and the mold 10 based on variations between RLT shot areas. For example, the CPU 40 (control unit) may control the pressing force or pressing amount of the mold 10. Specifically, in S260, the CPU 40 (control unit) issues a control command to increase the pressing amount of the mold lifting / lowering actuator 15 for a shot region having a large RLT. On the other hand, when the RLT is small, the CPU 40 (control unit) issues a control command to reduce the pressing amount of the mold lifting / lowering actuator 15 and adjusts so that the variation in RLT is reduced. The mold raising / lowering actuator 15 is a drive part which makes the mold 10 and a resist contact. For adjustment of RLT, the drive amount of the fine movement stage 3 in the Z direction may be controlled or both may be controlled instead of controlling the mold lifting / lowering actuator 15.

[Fifth Embodiment]
So far, it has been described that the change in RLT and HT is equivalent in the resist pattern shape, and that HT is not limited and only RLT is limited in the first polarization state. However, HT can be said to be a value determined to some extent by the height (depth) of the template (mold) due to the nature of the imprint process, and it can be said that the value does not vary so much. Therefore, HT can be limited to some extent assuming that the variation of HT is in the vicinity of a value determined by the height of the template. That is, for example, in FIG. 5C, since the number of pattern height candidates can be limited to three or less, the RLT range can also be limited. Therefore, the RLT is measured with high accuracy within a more limited range. be able to.
[Sixth Embodiment]
In the first embodiment, RLT is measured in the first polarization state (45-degree polarization) to narrow down the range of RLT, and the above-described narrowing is performed among a plurality of RLTs measured in the second polarization state (0-degree polarization). The method of determining RLT that matches the range of RLT is described. In this embodiment, in the process of FIG. 1 for determining the RLT described in the first embodiment, the wavelength that is a variable in the two polarization states is optimized, and measurement is performed using the optimized wavelength. Features. The optimized wavelength adds the noise that may occur during measurement to the reflected light intensity obtained by simulation, and the relationship between the intensity I (λ) and RLT value of different wavelengths is regressed. It is determined by a combination of variables that satisfy the allowable conditions and improve the prediction accuracy. Since the change in reflected light intensity obtained from the minute change in the concavo-convex pattern is minute, the measurement value is easily affected by noise. With this method, the parameters used for measurement are optimized, and high-precision measurement is possible by performing measurement that is robust against noise.

  The optimum variable determination method and measurement method of this embodiment will be described in detail with reference to FIGS. FIG. 13 is a flowchart showing the measurement method of this embodiment. First, in S300, the concave / convex pattern model defined on the simulator of the vector diffraction model and the variable λ used for creating the regression equation from the spectral characteristics of the reflected light of the concave / convex pattern model satisfy the conditions described in the first embodiment described above. Select one. The uneven pattern and spectral characteristics used to create the regression equation are obtained by changing a plurality of RLTs and HTs from the uneven pattern as a design value and using the spectral characteristics obtained for each shape. In the present embodiment, the spectral characteristic refers to the intensity of reflected light obtained for each wavelength. Next, a variable λ used for creating a regression equation is selected from the spectral characteristics obtained by the calculation. That is, a plurality of wavelengths λai (i = 1, 2,...) Having a large difference between the intensity change due to the RLT change and the intensity change due to the HT change and a plurality of wavelengths λb having the largest intensity change with respect to the RLT change are selected. In the present embodiment, the plurality of wavelengths λai are referred to as a first wavelength group, and the wavelength λb is referred to as a second wavelength. The reason for using the above two conditions is described in the first embodiment described above. In the present embodiment, in order to simplify the description, the case where there are two variables I (λ) will be described. That is, one each of the first wavelength and the second wavelength is selected. At this time, the second wavelength is fixed, and a plurality of the first wavelengths are selected.

Next, in S310, the magnitude of error (noise) of the measuring instrument used for measurement is obtained. As the types of noise at the time of measurement, random noise n1 that does not depend on the intensity of reflected light, such as dark noise of a CCD that is a detector, and noise n2 that is proportional to the intensity caused by a change in the output of the light source, etc. are considered. It is done. The magnitude of the noise n1 can be obtained from the output signal of the detector in a state where the light from the light source is not incident, for example. On the other hand, the magnitude of the noise n2 can be obtained, for example, by irradiating light using a non-patterned Si substrate as a sample and detecting the reflected light repeatedly. In other words, the change in the light source output can be measured by integrating the detected plurality of spectral characteristics between the wavelengths and removing the random noise n1 by averaging.
Next, in S320, a regression equation of the first wavelength group and second wavelength selected in S300 and RLT is created. As described above, since the first wavelength group is composed of a plurality of wavelengths, there are a plurality of combinations created by the first wavelength group and the second wavelength. A regression equation is created for each combination. In the present embodiment, the plurality of combinations are referred to as a combination group, and the plurality of regression equations are referred to as a regression equation group. Hereinafter, a method of creating a regression equation will be described with reference to FIG. FIG. 14 schematically shows an example of the combination group selected in step S300. As shown in FIG. 14, by using the calculated value of the intensity of the reflected light when using the light of the first wavelength λa1 and the second wavelength λb for each shape in which a plurality of RLTs and HTs are changed, An RLT regression equation is created with the calculated value of the intensity of reflected light at the second wavelength as a variable. As shown in FIG. 14, the regression equation uses the thickness of the recess as a dependent variable and the intensity of reflected light as an independent variable. Similarly to the above, regression equations are created for all combinations obtained in S300.

  Next, in S330, the noises n1 and n2 of the measuring instrument obtained in S310 are added to each of the regression equation groups created in S320 to determine a robust wavelength combination. Hereinafter, an optimal combination determination method will be described with reference to FIGS. 15A to 15C. In this embodiment, the second wavelength λb is fixed and the first wavelength λa is changed, and the evaluation is performed for ten combinations. FIG. 15A is a graph showing the residual (3σ) of the regression equation. FIG. 15B is a graph showing the reproducibility (3σ) when noise n1 is added and the average value of the difference between the predicted value and the true value. FIG. 15C is a graph showing the reproducibility (3σ) when noise n2 is added and the average value of the difference between the predicted value and the true value.

  Even when the conditions described in the first embodiment are used as shown in FIGS. 15A to 15C, there are large differences in measured values depending on combinations. From the result shown in FIG. 15A, for example, when the conditions of combination 4 and combination 5 are compared, the residuals are comparable. However, as shown in FIGS. 15B and 15C, it can be seen that the combination 4 is more robust against noise. In addition, although the robustness against the random noise n1 is high, there may be a combination 2 in which the residual of the regression equation becomes large. Therefore, a condition that increases the residual of the regression equation is not selected. In this way, an optimum parameter can be selected by evaluating a plurality of combinations and considering the influence of noise. The method for determining the optimum variable has been described above. In S340, the RLT is calculated using the optimum combination of variables obtained in S330.

  As described above, RLT can be measured with high accuracy by optimizing the wavelength used for measurement. In the present embodiment, two variables used for creating the regression equation are described. However, the present invention is not limited to this, and the same applies when there are three or more variables. Moreover, although the variable used for regression equation preparation was made into the wavelength, when incident light is made into a single wavelength and it changes by changing an incident angle, an incident angle can become a variable. The wavelength and incident angle optimized in the sixth embodiment are stored in the library in association with the shape of the concavo-convex pattern.

[Product Manufacturing Method]
A method of manufacturing a device (semiconductor integrated circuit element, liquid crystal display element, etc.) as an article includes a step of transferring (forming) a pattern onto a substrate (substrate, glass plate, film substrate, etc.) using the above-described imprint apparatus. Including. Further, the method may include a step of etching the substrate to which the pattern has been transferred. When manufacturing other articles such as patterned media (recording media) and optical elements, other processing steps for processing the substrate to which the pattern has been transferred may be included instead of the etching step.

Claims (7)

The pattern using the measurement result obtained by measuring the intensity of reflected light obtained from the pattern with light obliquely incident on the pattern, and information indicating the relationship between the intensity of the reflected light and the shape of the pattern A method of determining the incident condition of the light used to determine the thickness of the recess of
The measurement accuracy of the measuring instrument with respect to the thickness of the recess measured using the reflected light in the first polarization state from the pattern is set as the first accuracy, and the measurement accuracy is measured using the reflected light in the first polarization state. The measurement accuracy of the measuring device with respect to the difference thickness corresponding to the difference between the thickness of the convex portion of the pattern and the thickness of the concave portion is set as the second accuracy, and the measurement is performed using the reflected light in the second polarization state from the pattern. The measurement accuracy of the measuring instrument with respect to the thickness of the recess is the third accuracy, and the measurement accuracy of the measuring instrument with respect to the differential thickness measured using the reflected light in the second polarization state is the fourth accuracy, The measurement accuracy improves in the order of the second accuracy, the first accuracy, and the third accuracy, and the difference between the first accuracy and the second accuracy is greater than the difference between the third accuracy and the fourth accuracy. The first polarization state and the second polarization state as large as possible; Set,
One of a plurality of first incident conditions as the incident condition for generating the reflected light in the first polarization state and a plurality of first conditions as the incident condition for generating the reflected light in the second polarization state. For each of a plurality of combinations in combination with one of the two incident conditions, the regression equation is obtained using the obtained reflected light intensity as an independent variable and the thickness of the recess obtained using the information as a dependent variable. The first incidence condition corresponding to the regression equation that satisfies the tolerance of the regression error among the plurality of created regression equations is used as the incident condition for generating the reflected light in the first polarization state, and the regression equation is supported. And determining a second incident condition as an incident condition for generating reflected light in the second polarization state.   The method according to claim 1, wherein the intensity of reflected light as the independent variable is obtained by simulation based on an error of the measuring instrument. A method for generating a database including the information, comprising:
The method according to claim 1, wherein the first incident condition and the second incident condition determined by the method according to claim 1 or 2 are stored in the database in association with the information. A method of determining a first incident condition and a second incident condition by the method according to claim 1 or 2, and measuring a thickness of a concave portion of a pattern,
A first step of measuring the intensity of the reflected light in the first polarization state under the determined first incident condition;
A second step of measuring the intensity of reflected light in the second polarization state under the determined second incident condition;
A third step of determining the thickness of the concave portion based on the measurement results of the first step and the second step and the information;
A method characterized by comprising: The pattern is a resist pattern formed on a substrate by imprinting,
The method according to claim 4, wherein the first to third steps are performed at a plurality of locations on the substrate. An imprint apparatus for forming a resist pattern on a substrate,
A drive unit for contacting the resist with a mold for molding the resist;
A supply unit for supplying the resist to the substrate;
A control unit for controlling the operation of at least one of the drive unit and the supply unit so that the variation in thickness of the concave portion at the plurality of locations obtained by the method according to claim 5 is reduced;
An imprint apparatus comprising: Forming a pattern on a substrate using the imprint apparatus according to claim 6;
Processing the substrate on which the pattern is formed in the step;
A method for producing an article comprising:

Images